Learn Before
Strategizing from a Basic Model
Case context: You have built an initial, basic spam classification system. It is currently making many errors and is far from the performance you eventually want to achieve. Your manager suggests discarding it entirely and spending the next six months researching a state-of-the-art complex model from scratch.
Question: Based on Andrew Ng's advice, what should you do instead with the current basic system, and what benefit will this provide?
Sample answer: Instead of discarding it, you should examine how the current basic system functions. By analyzing its errors and behavior, you will quickly find clues that show the most promising directions in which to invest your time, potentially saving months or years of misdirected development.
Key points:
- Do not discard the basic system.
- Examine how the basic system functions.
- Use the examination to find clues for promising directions.
- This approach saves development time.
Rubric: The response should advise examining the basic system rather than discarding it, and state that this will reveal clues or promising directions.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI